Foundational to any implementation effort is understanding factors that promote or hinder change. Promoting factors are called ‘facilitators’ and hindering factors are ‘barriers.’ Determinants of these factors may be individual, such as the preferences, needs, attitudes, and knowledge of HCP, hospital leaders, patients, and visitors. An individual may be a strong, engaged leader (a facilitator) or an unengaged obstructor (barrier). Determinants may include a team’s composition or ways of communicating, an organization’s culture and capacity, or a system’s policies and resources.¹⁶ Organizationally, implementation may be facilitated or impeded by expectations and allocation of time (eg, competing priorities, data collection burden, provision of time to dedicate to an effort, fast turnaround at the expense of sustained processes), resources (eg, ease of adapting the EMR, staff capacity, and turnover), and leadership support¹⁷ or follower buy-in.¹⁸

Facilitators and barriers affect implementation to differing degrees. For example, an individual practitioner may oppose a change (ie, be a barrier), but the supervisor may be able to facilitate to overcome the opposition. Alternatively, a practitioner may champion a change, but without the support of the leadership, they may be unable to initiate it. Additional influential factors include context, level of engagement, and reliability (see Table 1 for a glossary of terms).

Failure by HCP to adhere to a guideline or standard is a common basis for initiating an improvement project. The Cabana Framework¹⁹ is a useful tool to understand how addressing real or perceived barriers can make an implementation effort successful. The framework employs 3 domains (ie, knowledge, attitude, and behavior) to understand the spectrum of barriers. The Expert Recommendations for Implementing Change (ERIC)²⁰ is another resource to help map barriers to strategies and identify appropriate implementation models or frameworks.

Data are essential for implementation to establish baselines, identify opportunities, measure progress, and justify use of resources to organizational leaders. No single method or measure will work for all situations, and standardized measures often are not available. Different frameworks lend themselves to specific methodologies, but any chosen method must do the following:

There are 3 general types of measures employed in implementation²¹:

Ideally, all 3 types of measures are included in a project. For example, a project seeking to reduce ventilator-associated pneumonia (VAP, ie, outcome) seeks to increase early extubation (ie, process) but needs to ensure a rise in reintubations or unplanned intubations is not occurring (ie, balancing). At times, a balancing measure may be difficult to identify due to the rarity of an event or an indirect relationship between outcome or process and balancing measures. In the VAP example, using the rate of nonventilator hospital-acquired pneumonia (NV-HAP) could help identify patients who develop NV-HAP following extubation (implying they were extubated too early), but not every patient with NV-HAP will have been intubated prior to diagnosis.

Similarly, choosing an outcome measure may be difficult if the likelihood of an outcome is multifactorial or exceedingly rare. In the case of hand hygiene, improved adherence (process) should prevent nosocomial transmission (outcome). However, an overall HAI rate may not reflect the change because hand hygiene is one of many potential factors that affect nosocomial transmission. Also, it may not be possible to count all prevented transmissions. For example, a patient would not be counted if they experienced onset of an upper respiratory infection following discharge but did not require readmission. In the case of antimicrobial resistance, improved contact precaution adherence for multidrug-resistant organism (MDRO) patients (ie, process) and improved antimicrobial stewardship (ie, process) should decrease morbidity and mortality due to antimicrobial resistance (ie, outcome) but may be difficult to demonstrate in a single center or short period.¹⁰ In these examples, the focus of the project might be the process and balancing measures, with attention to but not reliance on the outcome.

Often, measures that are standardized and utilized broadly are referred to as ‘benchmarks.’ Measures also may be developed locally and used in a combination with benchmark measures. For instance, a facility may start with NHSN event definitions²² and adapt them as definitions change over time or as needed based on the suitability to their setting (eg, pediatric, long-term care, home healthcare).²³ As another example, facilities typically use theWHO 5 Moments process measures to identify occasions when HCP should perform hand hygiene during patient care, but the methods of measurement of adherence can vary. A facility may measure adherence to all moments, adherence to a specific moment, or the amount of hand hygiene product used.²⁴ When possible, it is important to use the least resource-intensive means of data collection because resources are needed to feed data back to those who were monitored. Monitoring in combination with feedback has been shown to influence change and be more effective when delivered at a high frequency.²⁵

Table 2 provides a nonexhaustive list of methodologies commonly used for implementation measurement. For a research-focused overview, readers are encouraged to review the 2016 SHEA series on research methods in healthcare epidemiology and antimicrobial stewardship²⁶ and Livorsi et al.¹⁰

Pronovost et al articulated the 4 Es (Engage, Educate, Execute, and Evaluate), which may be the most pervasive model used in US healthcare epidemiology.³⁰ This model is well suited for large-scale projects that include multiple sites. Its cyclical nature allows for formative work and feedback to drive modifications and adaptations, and it provides a guide for resolving knowledge gaps through education. However, it does not include targeted strategies to address multilevel barriers that hinder putting knowledge into practice.

The following 4 E strategies guide organizational change efforts:

The 4Es guide improvement teams in planning to address key partners for the implementation process: senior hospital leaders, improvement team leaders, and frontline staff. Planning for and utilization of multifaceted interventions that address the 4Es, coupled with explicit efforts to improve teamwork and safety culture,³¹ have been associated with reductions in HAIs^32–34 and mortality³⁵ and increased cost savings.³⁶

The Behavior Change Wheel (BCW) is the result of an effort to link interventions with targeted behaviors more directly. It was developed by Michie et al,³⁷ who evaluated 19 existing behavior change frameworks for comprehensiveness (ie, applicability to any intervention), coherence, and link to a behavioral model. The result was a 3-layered tool:

One strength of the model is its nonlinearity, meaning that >1 behavioral system component, intervention function, and policy category can apply to an effort to affect change. Additionally, the model attempts to incorporate contextual influences on behavior, which the authors refer to as ‘automatic’ functions such as emotions and impulses that arise from associative learning and/or innate dispositions, as opposed to more reflective processes involving evaluations and plans. The BCW has been used widely in health promotion efforts such as smoking cessation³⁸ and obesity and sedentary behavior reduction,^39–41 and COM-B has been used to investigate hand hygiene adherence^42,43 and antibiotic prescribing.^44,45

CUSP focuses broadly on the idea of safety culture by empowering HCP to take responsibility for safety in their area, rather than defining specific domains of impact. As described by Pronovost et al⁴⁶ at Johns Hopkins, who developed and validated the CUSP model in intensive care settings in 2005, the program is composed of 8 steps:

The US Agency for Healthcare Research & Quality (AHRQ), the federal agency that provided funding for the development of CUSP, maintains an updated version of this framework on its website.^33,47 The AHRQ also funded “On the CUSP: Stop CAUTI,” a national program to reduce the incidence of CAUTI through technical and hospital culture adaptations rooted in the CUSP model.⁴⁸ This program focused on culture change.⁴⁹ CUSP has also been used in the ambulatory setting, specifically in the AHRQ “Safety Program for Improving Antibiotic Use,” a program derived from CUSP model concepts, designed to reduce overprescribing of antibiotics in primary care.⁵⁰ The AHRQ has further extended CUSP into ambulatory surgery, providing a full toolkit for the prevention of surgical-site infection on their website.⁵¹ Investigators and implementation scientists have continued to use the CUSP approach in a variety of clinical settings with mixed results. Although CUSP has shown success in preventing HAIs, such as CLABSI in US intensive care units^32,52 and CAUTI on medical-surgical floors in acute-care hospitals⁵³ and in nursing homes,⁵⁴ not all interventions using a CUSP-based approach have been successful.⁵⁵

The European and Mixed-Methods framework derives from the CFIR²⁷ and originated as the ‘InDepth’ work package,⁵⁶ a longitudinal qualitative comparative case study within the Prevention of Hospital Infections by Intervention and Training (PROHIBIT) study.^57,58 Specifically, InDepth sought to identify the role contextual factors play in barriers and facilitators to successful implementation.⁵⁹ The framework defines 3 qualitative measures of implementation success:

The framework has not been applied beyond the PROHIBIT outcomes of CLABSI rates and hand hygiene adherence,^57,58 but the reported results may be used to inform other approaches.

Getting to Outcomes (GTO)^® is a means of planning, implementing, and evaluating programs and initiatives developed for community settings. GTO^® seeks to build capacity for self-efficacy, attitudes, and behaviors to yield effective prevention practices.⁶⁰ The process involves 10 steps:

GTO^® has been utilized for numerous community-based initiatives, such as evidence-based sexual health promotion,⁶¹ a dual-disorder treatment program for veterans,⁶² and development of casework models for child welfare services.⁶³ Additionally, the RAND Corporation has published a guide to develop community emergency preparedness programs. This guide breaks down each of the 10 GTO^® steps, provides materials and examples,⁶⁴ and may facilitate the use of GTO^® in implementing infection prevention interventions.

The Model for Improvement⁶⁵ was developed by the Associates for Process Improvement based on Deming’s System of Profound Knowledge.⁶⁶ It has since been adopted widely, perhaps most notably by the Institute for Healthcare Improvement (IHI) in its 100,000 and 5 Million Lives campaigns of the early 2000s.⁶⁷ The model has been used to accelerate change in a variety of healthcare and public health settings,^68–70 and subspecialists have created primers focused on their practice areas.^71–74 The Model for Improvement begins with 3 questions:

What are we trying to accomplish?

Reach, effectiveness, adoption, implementation, and maintenance make up the 5 dimensions of the planning and evaluation framework RE-AIM, developed to address the failures and delays in translating scientific evidence into policy and practice.^75,76 By utilizing these 5 dimensions at individual and ecological levels, teams can better understand the effectiveness of programs as they are implemented in real-world community settings.^77,78

RE-AIM is useful for planning an intervention, the outcomes that will be measured, and evaluating whether the intervention has met its goals.⁷⁹ All 5 dimensions are not always addressed. In recent years there has been greater emphasis on pragmatic application of the framework to determine which dimensions an organization should prioritize.⁸⁰ It also provides ideas for quantitative measurements of outcomes.

RE-AIM was utilized to evaluate an antimicrobial stewardship program in an ICU in South Africa,⁸¹ dissemination and implementation of clinical practice guidelines for sexually transmitted infections,⁸² and promotion of vaccination via digital technology.⁸³ Recently, to better understand the implementation of contact tracing for emerging infectious diseases, RE-AIM was used to evaluate individual and systems-level predictors of success of an emergency volunteer COVID-19 contact tracing program in Connecticut.^76,78,84 Investigators concluded that the program fell short of CDC benchmarks for time and yield, largely due to difficulty collecting the information necessary for outreach.

The Replicating Effective Programs (REP) framework may be used to balance needs of the target population with the core elements of successfully implemented interventions⁸⁵ and to maximize fidelity to core interventions that have been rigorously tested and have produced statistically significant positive results.⁸⁶

There are 4 phases of REP⁸⁷:

REP may be useful when adapting interventions for a specified target audience within healthcare. It also may be applied across the continuum of care (eg, acute care to long-term care) or in multifacility systems when local institutional culture dictate the need for adaptation. When used to disseminate evidence-based HIV prevention interventions to community-based organizations, the application of REP to packaging, HCP training, and technical assistance resulted in more effective uptake than dissemination alone.⁸⁸

The Theoretical Domains Framework (TDF) was initially developed to conduct research on the behavior of HCP as it relates to implementing evidence-based practices.⁸⁹ The initial organization of TDF⁹⁰ was modified following a formal validation exercise,⁹¹ which yielded 14 domains to identify relevant cognitive, affective, social, and environmental influences on behavior.⁸⁹ TDF has been used widely to understand and influence HCP,⁹² patient, and population behaviors, most commonly with qualitative methods (eg, surveys, interviews, and/or focus groups).⁸⁹ One salient example is a patient safety effort to properly place nasogastric tubes. A team utilized a validated TDF-based questionnaire to identify relevant domains that were then explored in focus groups to help connect theory to techniques to change behavior.⁹³ More recently, TDF was used to develop the Choosing Wisely De-Implementation Framework⁹⁴ that proposes to reduce low-value care, defined as a test or treatment for which there is no evidence of patient benefit or where there is evidence of more harm than benefit.⁹⁵ A guide to TDF use⁸⁹ and a 4-step systematic approach to using TDF⁹⁶ were published to help teams design and follow through with an intervention. TDF has been linked to the COM-B model (used in the aforementioned BCW) and has been used in combination with other frameworks when the time necessary to complete interviews and focus groups was limited.⁹⁷

Allthough national implementation studies have succeeded in preventing several different HAIs, investigators have not seen the same success when focused on facilities most in need of help—hospitals underperforming with respect to HAI prevention. The CDC-funded national prospective, interventional, quality improvement program, CDC STRIVE (States Targeting Reduction in Infections Via Engagement), focused on reducing CLABSIs, CAUTIs, Clostridioides difficile infections, and methicillin-resistant Staphylococcus aureus (MRSA) bloodstream infections in hospitals with a disproportionately high burden of HAIs.⁹⁸ Although nearly 400 US hospitals participated in this multimodal, multifaceted, partner-facilitated program, they did not see significantly reduced rates of CLABSI,⁹⁹ CAUTI,¹⁰⁰ C difficile infection,¹⁰¹ or MRSA bloodstream infection.¹⁰² More recently, the Agency for Healthcare Research and Quality (AHRQ) funded a national program that invited US hospitals that had at least 1 adult ICU with elevated CLABSI or CAUTI rates to participate in an externally facilitated program implemented by a national project team and state hospital associations using the Comprehensive Unit-based Safety Program (CUSP) framework.¹⁰³ Results from the first 2 cohorts (366 recruited ICUs from 220 hospitals in 16 states and Puerto Rico) revealed no statistically significant reductions in CLABSI, CAUTI, or catheter utilization in the 280 ICUs that completed the program.¹⁰³ These researchers cite a number of possible factors contributing to the disappointing result, including underutilization of training and coaching resources, lack of infrastructure, and a different selection process for participants (eg, identifying low-performing units or those that had been unsuccessful to date versus asking for volunteers for earlier CUSP work, which may have selected for early adopters and high performers). Investigations in this and similar cohorts could help elucidate why hospitals with disproportionately high HAI rates have not yet seen significant reductions in HAIs despite broad-based efforts. Increased focus on the development, adaptation, and utilization of implementation models and frameworks to infection prevention and control may help identify implementation gaps that contribute to lack of improvement and guide their closure.

A long-term goal of any implementation effort is to sustain and advance short-term gains. Ideally, sustaining gains occurs with less intensity than initial efforts, maintaining gains or improving at a slower rate and allowing resources to be directed to another effort. Characteristics of successfully sustained interventions have included those that are incorporated into the standard workflow, have effective champions to shepherd the effort and re-engage when necessary, can be modified over time, fit with an organization’s mission and procedures, provide easily perceived benefits to staff members and/or clients, and are supported by partner organizations.¹⁰⁴ It can be difficult to meet those criteria in healthcare, where changes in workflows and staff are frequent.¹⁰⁵ Demonstrating successfully sustained implementation should include evidence of (1) sustainment, that is, sustained use of an evidence-based intervention (process measure), and (2) sustainability, that is, sustained benefits of an evidence-based intervention (outcome measure).¹⁰⁶

Linking ongoing process data to ongoing outcome data can prove challenging. In one study, CLABSI reduction in ICUs was sustained for a decade, but process measurement was not performed.³⁴ A study on CAUTI prevention at a Veterans’ Affairs (VA) hospital found that, 8 years after implementation, appropriateness of urinary catheters remained high and stable.¹⁰⁵ Catheter use decreased, but the facility was unable to report outcome data.¹⁰⁵ These researchers hypothesized that success was due to a 3-component, evidence-based intervention, institutionalization of the interventions (ie, standardizing nursing assessments and handoffs that included the intervention), and effective champions who re-engaged when necessary. Additionally, a study on hand hygiene on 2 hospital units in Italy found that adherence dropped after 4 years (from 84.2% to 71%) despite maintaining champions and processes.¹⁰⁷ Recent proposals for standard definitions^106,108 and modified ERIC strategies to account for sustainment and identify interventions that yield short-term and sustained improvements¹⁰⁹ can form a basis for future research and understanding.

It is increasingly evident that implementation is essential to ensuring that evidence-based interventions are performed to generate desired outcomes and to meet infection prevention and control and antimicrobial stewardship goals.¹⁰ Furthermore, a detailed implementation plan in a specific healthcare setting for a given intervention is necessary for success, as the implementation approach in one facility may not be reproducible, with the desired effect, in another. In this article we have provided an overview of implementation in a general sense, with a glossary of terms, broader discussion of key methods, models, and frameworks, possible future areas of study, as well as links to resources readers can use to initiate or continue their implementation journey.